The invention relate to a device for the monitoring and prognosis of the failure probability of inductive proximity sensors for monitoring of the position of movable switch rails or rail components, in which the proximity sensor has at least one coil supplied by an oscillator, and the sensor current flowing by means of variable attenuation is measured and then fed to an evaluation circuit.
For the monitoring of the assembly, or disassembly of tongues on jaw rails, inductive sensors may be used, for example. In order to grasp the assembly of the tongue on a jaw rail, the inductive sensor may be attached in the rack of the jaw rail, whereby either the non-attenuated condition of this sensor, or an additional, specifically designed inductive sensor may be used for the assembly of the tongue. Such sensors deliver a certain current depending on the attenuation, and the current reception of the sensor can be monitored, and the distance information can be gained in this way. A method for the monitoring of the condition of switch rails is known from AT 399 851 B, in which additional signals gained during the riding of the switch rails are evaluated, and the smallest reading of each distance is stored, whereby the change of the smallest reading measured, as well as at least a first threshold reading for the smallest distance are compared to one another. A premature wear in the area of the tongue rail of a switch rail should be able to be recognized in this way. It was recommended in WO 97/33784 to design the sensor as a continuous proximity sensor, and to perform two separate evaluations, whereby one evaluation is to yield each distance, and the second evaluation a functional control of the sensor, whereby the predetermined tolerance windows of the characteristic line are considered the standard.
Generally, the sensors designed as continuous proximity sensors are mostly designed as inductive sensors, whereby a coil with a freely oscillating oscillator is used, the resonant amplitude of which changes with the proximity of ferromagnetic, or electrically conductive material, whereby a corresponding change of the current consumption can be measured. The corresponding proximity information can be gained in an evaluation circuit from the measured currency consumption.
In order to monitor the correct function of inductive proximity sensors, it has already been recommended in DE-C-1 50 212 to use test signals. By means of such test signals, an electric attenuation, or de-attenuation should be achieved, whereby the evaluation of the change in the signal with engaged test signal, and without test signal should result in an evaluation of the availability of the proximity sensor. With such circuit arrangements it can be principally recognized whether a sensor was correctly connected, and particularly, whether a sensor is defect, as the engagement of a test signal does not show any changes in this case, that can be evaluated.
The invention aims to create a device for the monitoring and the prognosis of the failure probability of inductive proximity sensors for the monitoring of the position of movable switch rails or rail components, in which the principle of the known test attenuation is used, but whereby a self test device is to be created for inductive proximity sensors for the entire signal behavior, in order to generate a failure prognosis for the sensor from the changes of a possible characteristic line course. Possible causes for an increased failure probability are particularly a decrease in the isolation resistance between the conductors by water penetration, and therefore a formation of parasitic resistances, mechanical damage, or damage to the assembly components in the circuit, as well as errors in the sensor electronics, which lead to a change in the proximity current characteristic line, whereby the wrong proximity information would be gathered. The inventive device should show such changes in a timely fashion so that the maintenance, or the exchange of a sensor can occur long before its actual failure. In order to solve this task, the inventive device essentially consists of the fact that characteristic lines of the sensor are stored in the electric not additionally attenuated condition, and in the electric additionally attenuated condition for the course of the sensor currents in dependency of the distance of the movable switch rails, or rail components, i.e. the mechanical attenuation, and that the measuring currents corresponding to the mechanical attenuation condition, as well as to the respective additionally electric attenuated condition are cyclic scanned, and the respective measuring currents, or measurement reading pairs are fed to a comparison and evaluation circuit, in which the differences due to the characteristic line are compared to the measured differences. Inductive sensors are usually removed after the adjustment, whereby a characteristic line of the sensor is received for the various positions of the tongue relative to the jaw rail. Due to he fact that this characteristic line, as well as an additional characteristic line, in which the sensor was additionally electrically attenuated, are received, and both of these characteristic lines are stored, it is subsequently possible to make predictions for the failure probability independently of the respective position of the tongue relative to the jaw rail. Normally, a defective sensor in a non-attenuated condition will have a clearly distinguishable signal as opposed to an intact sensor. In a maximum mechanically non-attenuated condition of such a sensor it has been shown, however, that the signals of a defective sensor essentially cannot be differentiated from the signals of an intact sensor. Not until the actual checking of the measurement readings for an electrically attenuated circuit arrangements can the differences between intact sensors and defective sensors be clearly recognized, if a respective comparison with the original characteristic lines of the intact sensor is performed for this purpose. For this purpose, the invention suggests that the measuring currents corresponding to a mechanical attenuation condition, as well as the corresponding additional electrically attenuated measuring currents are cyclic scanned and the respective measuring currents, or the measurement reading pairs are fed to a comparison and evaluation circuit. In such a comparison and evaluation circuit, an evaluation can subsequently be performed by means of the stored characteristic lines, whereby the differences to the differences due to the characteristic lines of the intact sensors enable a corresponding prognosis of the failure probability.
In a particularly advantageous way, the embodiment is chosen so that the signals of the measurement reading pairs for at least to different mechanical attenuations are fed to the evaluation circuit. It was shown across the entire signal course that the difference of the measurement readings for an intact sensor and a defective sensor in the electrically non-attenuated condition depends on the actual position, and thereby on the distance of the measurement object. It was particularly shown that this difference is larger with maximum mechanical attenuation, than with minimum mechanical attenuation, i.e. with a disassembled tongue. This behavior of the characteristic line of the intact, and that of a defective sensor without electric attenuation directly results in the fact that exact statements on the extent of a defect cannot be made across a large partial area of the characteristic line, as the difference between the signals of an intact and that of a defective sensor is not significant. Significant changes only appear in the maximum mechanically attenuated position, however whereby here, without the assistance of the characteristic line of an electrically attenuated sensor, the wrong proximity readings could be achieved, as particularly a measuring current that is too low would already signalize an attachment position already at a distance to the attachment, and corresponds to the measuring current that corresponds to the attachment of a tongue on a jaw rail in an intact sensor. Not until the comparison with the measurement readings and the characteristic lines of an additional electrically attenuated sensor, are such differences able to be evaluated, and allow the corresponding conclusions to the errors and the failure probability.
In a particularly simple way, the electric attenuation may occur by means of an increase in supply voltage. This is especially recommended for the use of the so-called two-wire technology, in which a statistical decrease of the failure probability occurs due to the decrease of the amount of electric wires. The circuit for the determination of the additional electric attenuated signal can be advantageously designed so that the electric attenuation occurs by means of a tap of the sensor coil, whereby a transistor is advantageously intended that controls the electric attenuation by means of a resistance at the tap of the sensor coil, the base of which is connected to the voltage source by moans of a Z diode. By simply increasing the supply voltage, the direct switching in of the electric attenuation, and obtaining merely the measuring reading for an exclusive mechanical attenuation by means of decreasing the supply voltage can be achieved in a simple way. In an advantageous way it is proceeded in such a way that at least one evaluation of the difference of the measurement reading pairs occurs in a mechanically non-attenuated position, whereby the evaluation may occur in such a way that a respective constant current is measured at a respective distance.
By means of the cyclic switching in of a resistance to the resonant circuit, an attenuation of the resonant circuit defined in size therefore occurs, and by mews of the switching in by increasing the supply voltage, the reach may be found with merely two wires. Functionally seen, the attenuation by the sensor housing, assembly and measurement subjects, is the loss of eddy currents that lead to a decrease of the resonant amplitude of the oscillator as losses of effect, and therefore to a decrease of current consumption.
The effect of a resistance parallel to the resonant circuit equals the affect of eddy current losses, and almost every sensor error leading to a change in the signal behavior can be determined by means of a cyclic test attenuation, and a corresponding evaluation of the sensor current in this way.
By means of making a comparison with the corresponding characteristic lines, a prognosis may also be achieved with regard to the sensor behavior in the attenuation by the measurement object, i.e. in the attenuation by the tongue moving from the assembly to the disassembly. When such a prognosis does not achieve the expected sensor current, a sensor error may be the conclusion.